UNICEF Data Story: Global Child Vulnerability & DTP Vaccination

Spring 2025 BAA1030 Data Analytics & Story Telling (20074)

Student Name: Isha Tanwar

Student ID: 48614

Programme: MSc in Management (Strategy)

Thanks to Professor Dr. Damien Dupre, Dublin City University, for his unwavering guidance and support.

Code

import pandas as pd
from plotnine import *
from sklearn.preprocessing import MinMaxScaler
import plotly.graph_objects as go

# Load dataset
merged_df = pd.read_csv("merged_final.csv")

# Derived column
merged_df["OrphanRate_per_1000"] = (
    merged_df["OrphanCount"] / merged_df["Population, total"]
) * 1000

# Rename for convenience
merged_df = merged_df.rename(columns={
    "GDP per capita (constant 2015 US$)": "GDP",
    "Life expectancy at birth, total (years)": "LifeExpectancy"
})

Executive Summary

This report explores global disparities in child vulnerability using data from UNICEF and the World Bank.
It combines key indicators — GDP, immunization rates, orphan counts, and life expectancy — into a composite index.
Visual analytics highlight countries most at risk, helping align interventions with SDG 3 (Health) and SDG 10 (Inequality).

Introduction

This report provides a data-driven exploration of global child vulnerability, focusing on DTP (Diphtheria, Tetanus, and Pertussis) vaccination coverage. Using datasets from UNICEF and the World Bank, this analysis combines economic, health, and demographic indicators to expose inequities in child well-being.

Scatter Plot: GDP vs DTP Vaccination Coverage

Code

scatter_df = merged_df[["Country", "DTP", "GDP"]].dropna()

scatter_plot = (
    ggplot(scatter_df, aes(x="GDP", y="DTP")) +
    geom_point(color="green", alpha=0.7) +
    geom_smooth(method='lm', color="darkblue") +
    labs(
        title="GDP per Capita vs DTP Vaccination Coverage",
        x="GDP per Capita (USD)",
        y="DTP Coverage (%)"
    ) +
    theme_minimal()
)

scatter_plot.draw()

Insight: Countries with higher GDP per capita tend to achieve better DTP vaccination rates. This suggests economic capacity plays a pivotal role in health system performance and immunization reach.

Bar Chart: Top 10 Countries by Orphanhood

Code

import pandas as pd
from plotnine import *

# Get the most recent year in your dataset
latest_year = merged_df["Year"].max()

# Filter for top 10 countries by OrphanCount
bar_data = (
    merged_df[merged_df["Year"] == latest_year]
    .dropna(subset=["OrphanCount"])
    .sort_values(by="OrphanCount", ascending=False)
    .head(10)
)

# Format OrphanCount as readable labels like "13.6M"
def format_millions(val):
    return f"{val / 1_000_000:.1f}M"

bar_data["Label"] = bar_data["OrphanCount"].apply(format_millions)

# Plot with nice labels
bar_chart = (
    ggplot(bar_data, aes(x="reorder(Country, OrphanCount)", y="OrphanCount")) +
    geom_bar(stat="identity", fill="#f15b42", width=0.7) +
    geom_text(
        aes(label="Label"),
        format_string="{:}",
        nudge_y=2e5,  # small offset to move label slightly right
        size=11,
        color="black",
        ha="left"
    ) +
    coord_flip() +
    labs(
        title=f"Top 10 Countries by Orphanhood ({latest_year})",
        x="Country",
        y="Number of Orphaned Children"
    ) +
    theme_minimal() +
    theme(
        figure_size=(14, 8),
        axis_text=element_text(size=14),
        plot_title=element_text(size=18, weight="bold"),
        axis_title=element_text(size=15)
    )
)

bar_chart.draw()

Insight: Countries like Nigeria, DR Congo, and Pakistan top the orphanhood chart, reflecting the heavy burden of conflict, disease, and poverty.
This visualization underscores how systemic crises directly affect children’s lives — a call to accelerate child protection and family-strengthening programs in high-risk regions.

Time Series: Orphanhood Trends (Multi-Country Normalized)

Code

focus_countries = ["Ethiopia", "Nigeria", "Yemen", "Somalia", "Guinea", "Ukraine"]
ts_data = merged_df[
    merged_df["Country"].isin(focus_countries) &
    merged_df["OrphanRate_per_1000"].notna()
].sort_values(["Country", "Year"])

orphan_trend_plot = (
    ggplot(ts_data, aes(x="Year", y="OrphanRate_per_1000", color="Country")) +
    geom_line(size=1.2) +
    geom_point(size=2, alpha=0.8) +
    labs(
        title="Orphanhood Rate per 1000 Children",
        subtitle="Normalized by population",
        x="Year",
        y="Orphan Rate (per 1000)"
    ) +
    theme_minimal()
)

orphan_trend_plot.draw()

Insight: While absolute orphan counts vary, the normalized trends show that countries like Somalia and Yemen consistently exhibit high orphanhood rates per 1000 children. This reveals not just scale, but intensity of vulnerability, adjusted for population size.

Faceted View: Orphanhood Trends by Country

Code

facet_chart = (
    ggplot(ts_data, aes(x="Year", y="OrphanRate_per_1000")) +
    geom_line(color="#c778a2", size=1.2) +
    geom_point(color="#af4e88", size=2) +
    facet_wrap("~Country") +
    labs(
        title="Orphanhood Trends by Country (Faceted)",
        x="Year",
        y="Orphan Rate per 1000"
    ) +
    theme_minimal()
)

facet_chart.draw()

Insight: Country-specific plots make it easier to spot individual trends—e.g., a steady rise in Ukraine vs. fluctuations in Nigeria. This enables policymakers to tailor child protection strategies by context.

World Map: DTP Vaccination Coverage (Rotating Globe)

Code

globe_data = merged_df[merged_df["Year"] == latest_year][["Country", "DTP"]]

custom_colors = [
    [0.0, "#fce4ec"], [0.2, "#f8bbd0"], [0.4, "#f48fb1"],
    [0.6, "#ce93d8"], [0.8, "#ab47bc"], [1.0, "#6a1b9a"]
]

fig = go.Figure(
    data=go.Choropleth(
        locations=globe_data["Country"],
        locationmode="country names",
        z=globe_data["DTP"],
        colorscale=custom_colors,
        colorbar_title="DTP Coverage (%)",
        hovertext=globe_data["Country"],
        hoverinfo="text+z"
    )
)

fig.update_geos(
    projection_type="orthographic",
    showcoastlines=True,
    showland=True,
    showocean=True,
    coastlinecolor="gray",
    landcolor="white",
    oceancolor="lavender"
)

fig.update_layout(
    title_text=f"DTP Vaccination Coverage (Rotating Globe) – {latest_year}",
    height=700,
    geo=dict(showframe=False)
)

fig.show()

Insight: The globe highlights stark disparities: Western Europe and North America show high vaccination coverage, while parts of Sub-Saharan Africa and South Asia lag behind. These spatial gaps in access reflect systemic inequities that must be addressed to achieve SDG 3 (Health for All).

Code

indicators = merged_df[[
    "DTP", "GDP", "LifeExpectancy", "OrphanRate_per_1000"
]].dropna()

# Invert values so high = more vulnerable
inverted = indicators.copy()
inverted["DTP"] = 100 - inverted["DTP"]
inverted["GDP"] = inverted["GDP"].max() - inverted["GDP"]
inverted["LifeExpectancy"] = inverted["LifeExpectancy"].max() - inverted["LifeExpectancy"]

# Normalize and compute CVI
scaler = MinMaxScaler(feature_range=(0, 100))
normalized = scaler.fit_transform(inverted)
merged_df.loc[indicators.index, "Child Vulnerability Index"] = normalized.mean(axis=1)

# Reattach CVI to dataframe
merged_df.loc[indicators.index, "Child Vulnerability Index"] = normalized.mean(axis=1)

# Create CVI data subset for plotting
cvi_data = merged_df.dropna(subset=["Child Vulnerability Index"]).copy()
cvi_data = cvi_data.rename(columns={"Child Vulnerability Index": "CVI"})

Weighted Index: Bottom 10 by Vulnerability

Code

from plotnine import (
    ggplot, aes, geom_bar, geom_text, geom_hline, coord_flip, labs,
    theme_minimal, theme, element_text
)

# Step 1: Create CVI if not done yet
if "Child Vulnerability Index" not in merged_df.columns:
    indicators = merged_df[[
        "DTP", "GDP", "LifeExpectancy", "OrphanRate_per_1000"
    ]].dropna()
    inverted = indicators.copy()
    inverted["DTP"] = 100 - inverted["DTP"]
    inverted["GDP"] = inverted["GDP"].max() - inverted["GDP"]
    inverted["LifeExpectancy"] = inverted["LifeExpectancy"].max() - inverted["LifeExpectancy"]
    scaler = MinMaxScaler(feature_range=(0, 100))
    normalized = scaler.fit_transform(inverted)
    merged_df.loc[indicators.index, "Child Vulnerability Index"] = normalized.mean(axis=1)

# Step 2: Filter properly
cvi_data = merged_df.dropna(subset=["Child Vulnerability Index"]).copy()
cvi_data = cvi_data.rename(columns={"Child Vulnerability Index": "CVI"})

# ⚠️ Avoid duplicates — get only latest year if your dataset is multi-year
if "Year" in cvi_data.columns:
    latest_year = cvi_data["Year"].max()
    cvi_data = cvi_data[cvi_data["Year"] == latest_year]

# Step 3: Bottom 10 CVI scores
bottom_10 = cvi_data.sort_values("CVI", ascending=False).tail(10)
avg_cvi = round(cvi_data["CVI"].mean(), 2)

# Step 4: Beautiful CVI bar chart
bar_index = (
    ggplot(bottom_10, aes(x="reorder(Country, CVI)", y="CVI")) +
    geom_bar(stat="identity", fill="#d88fb3") +
    geom_text(aes(label="round(CVI, 1)"), format_string="{:.1f}", nudge_y=1, size=10, color="black") +
    geom_hline(yintercept=avg_cvi, linetype="dashed", color="gray") +
    coord_flip() +
    labs(
        title="Bottom 10 Countries by Child Vulnerability Index",
        x="Country",
        y="CVI (0–100)"
    ) +
    theme_minimal() +
    theme(
        figure_size=(12, 6),
        axis_text=element_text(size=11),
        plot_title=element_text(size=14, weight='bold')
    )
)

bar_index.draw()

Insight: The CVI shows which countries face the gravest challenges in child well-being. Factors like low GDP, low immunization, and high orphan rates converge to elevate vulnerability. The bottom 10 countries—many of them in Africa—need holistic interventions across health, economic, and social sectors.

Simulated DTP Coverage (based on GDP)

Code

import pandas as pd
import altair as alt
from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import MinMaxScaler

# 1. Load Data
df = pd.read_csv("merged_final.csv")

# 2. Rename Columns Properly
df = df.rename(columns={
    "GDP per capita (constant 2015 US$)": "GDP",
    "Life expectancy at birth, total (years)": "LifeExpectancy"
})

# 3. Filter for years 2014-2024
df = df[df["Year"].between(2014, 2024)]

# 4. Create OrphanRate
if "OrphanRate_per_1000" not in df.columns:
    df["OrphanRate_per_1000"] = (df["OrphanCount"] / df["Population, total"]) * 1000

# 5. Compute CVI
cvi_fields = ["DTP", "GDP", "LifeExpectancy", "OrphanRate_per_1000"]
cvi_df = df.dropna(subset=cvi_fields)[["Country"] + cvi_fields].copy()

inv = cvi_df[cvi_fields].copy()
inv["DTP"] = 100 - inv["DTP"]
inv["GDP"] = inv["GDP"].max() - inv["GDP"]
inv["LifeExpectancy"] = inv["LifeExpectancy"].max() - inv["LifeExpectancy"]

scaler = MinMaxScaler((0, 100))
normalized = scaler.fit_transform(inv)

cvi_df["CVI"] = normalized.mean(axis=1)

# 6. Bottom 10 countries by CVI
bottom_countries = (
    cvi_df.groupby("Country")["CVI"]
    .mean()
    .sort_values(ascending=False)
    .head(10)
    .index.tolist()
)

# 7. Filter latest year for these countries
latest_year = df["Year"].max()
latest_df = df[(df["Year"] == latest_year) & (df["Country"].isin(bottom_countries))]

# 8. Fit Linear Regression Model on GDP vs DTP
train_df = df.dropna(subset=["GDP", "DTP"])
model = LinearRegression()
model.fit(train_df[["GDP"]], train_df["DTP"])

# 9. Simulate 12% GDP increase
latest_df["GDP_Increased"] = latest_df["GDP"] * 1.12

# 10. ⚡ Drop NaN before Predicting
latest_df = latest_df.dropna(subset=["GDP_Increased"])

# 11. Predict Simulated DTP
gdp_for_prediction = latest_df[["GDP_Increased"]].rename(columns={"GDP_Increased": "GDP"})
latest_df["Simulated_DTP"] = model.predict(gdp_for_prediction)

# 12. Prepare data for Altair
plot_df = latest_df[["Country", "DTP", "Simulated_DTP"]].copy()
plot_df = plot_df.melt(id_vars="Country", var_name="Type", value_name="Coverage")

# 13. Horizontal Stacked Bar Chart
country_order = latest_df.sort_values("DTP")["Country"].tolist()

chart = (
    alt.Chart(plot_df)
    .mark_bar()
    .encode(
        y=alt.Y("Country:N", sort=country_order, title="Country"),
        x=alt.X("Coverage:Q", stack="zero", title="DTP Coverage (%)"),
        color=alt.Color(
            "Type:N",
            scale=alt.Scale(domain=["DTP", "Simulated_DTP"], range=["#e1a8c1", "#66bb6a"]),
            legend=alt.Legend(title="Coverage Type")
        ),
        tooltip=["Country", "Type", "Coverage"]
    )
    .properties(
        title={
            "text": "Simulated DTP Coverage Based on GDP (Policy Simulator)",
            "subtitle": "Actual vs Simulated DTP Coverage for Bottom 10 Vulnerable Countries",
            "fontSize": 18,
            "subtitleFontSize": 14
        },
        width=350,
        height=250
    )
)

chart

Insight: Countries such as Somalia, Guinea, and Nigeria demonstrate large potential gains in DTP coverage if GDP is increased moderately.
Simulated scenarios empower evidence-driven investments targeting the most vulnerable regions.

SDG Alignment & Policy Implications

SDG 3: Good Health and Well-being

 Investing in immunization helps build resilient health systems.

SDG 10: Reduced Inequalities

 Closing vaccination gaps supports vulnerable populations.

Conclusion

This analysis provides evidence of how economic and demographic conditions impact child vulnerability. Better policies rely on data-driven insights like these.

References

UNICEF Datasets. (2024). Child Vulnerability Indicators.
World Bank Open Data. (2024). Economic and Health Indicators by Country.
United Nations. (2023). Sustainable Development Goals (SDG) Framework.

Report prepared by Isha Tanwar | Spring 2025 | Dublin City University